From dafc84390db4b57e5958f1af386c778b481c7d04 Mon Sep 17 00:00:00 2001
From: Michael Vogel <icarus@dabo.de>
Date: Thu, 5 Jul 2012 23:36:00 +0200
Subject: [PATCH] Improvments of the HTML to make a better bb2markdown
 conversion (nested lists, youtube, vimeo, ...)

---
 include/api.php         |  1 +
 include/bb2diaspora.php |  4 ++--
 include/bbcode.php      | 46 +++++++++++++++++++++++++++++++++--------
 include/items.php       | 19 ++++++++++-------
 4 files changed, 52 insertions(+), 18 deletions(-)

diff --git a/include/api.php b/include/api.php
index d790b4b875..629b5fcc2a 100644
--- a/include/api.php
+++ b/include/api.php
@@ -1726,3 +1726,4 @@ notifications/leave
 blocks/exists
 blocks/blocking
 */
+
diff --git a/include/bb2diaspora.php b/include/bb2diaspora.php
index 77a5f5c2a0..4a82635e57 100644
--- a/include/bb2diaspora.php
+++ b/include/bb2diaspora.php
@@ -112,7 +112,7 @@ function bb2diaspora($Text,$preserve_nl = false) {
 	// Note that to get nested lists to work for Diaspora, we would need
 	// to define the closing tag for the list elements. So nested lists
 	// are going to be flattened out in Diaspora for now
-	$endlessloop = 0;
+/*	$endlessloop = 0;
 	while ((((strpos($Text, "[/list]") !== false) && (strpos($Text, "[list") !== false)) ||
 	       ((strpos($Text, "[/ol]") !== false) && (strpos($Text, "[ol]") !== false)) || 
 	       ((strpos($Text, "[/ul]") !== false) && (strpos($Text, "[ul]") !== false))) && (++$endlessloop < 20)) {
@@ -125,7 +125,7 @@ function bb2diaspora($Text,$preserve_nl = false) {
 		$Text = preg_replace_callback("/\[ul\](.*?)\[\/ul\]/is", 'diaspora_ul', $Text);
 		$Text = preg_replace_callback("/\[ol\](.*?)\[\/ol\]/is", 'diaspora_ol', $Text);
 	}
-
+*/
 	// Convert it to HTML - don't try oembed
 	$Text = bbcode($Text, $preserve_nl, false);
 
diff --git a/include/bbcode.php b/include/bbcode.php
index 9071c767b6..988e75d417 100644
--- a/include/bbcode.php
+++ b/include/bbcode.php
@@ -251,21 +251,25 @@ function bbcode($Text,$preserve_nl = false, $tryoembed = true) {
 	$Text = preg_replace("/\[img\](.*?)\[\/img\]/ism", '<img src="$1" alt="' . t('Image/photo') . '" />', $Text);
 
 
-	$Text = preg_replace("/\[video\](.*?\.(ogg|ogv|oga|ogm|webm|mp4))\[\/video\]/ism", '<video src="$1" controls="controls" width="425" height="350"><a href="$1">$1</a></video>', $Text);
-
-	$Text = preg_replace("/\[audio\](.*?\.(ogg|ogv|oga|ogm|webm|mp4|mp3))\[\/audio\]/ism", '<audio src="$1" controls="controls"><a href="$1">$1</a></audio>', $Text);
-
 	// Try to Oembed
 	if ($tryoembed) {
+		$Text = preg_replace("/\[video\](.*?\.(ogg|ogv|oga|ogm|webm|mp4))\[\/video\]/ism", '<video src="$1" controls="controls" width="425" height="350"><a href="$1">$1</a></video>', $Text);
+		$Text = preg_replace("/\[audio\](.*?\.(ogg|ogv|oga|ogm|webm|mp4|mp3))\[\/audio\]/ism", '<audio src="$1" controls="controls"><a href="$1">$1</a></audio>', $Text);
+
 		$Text = preg_replace_callback("/\[video\](.*?)\[\/video\]/ism", 'tryoembed', $Text);
 		$Text = preg_replace_callback("/\[audio\](.*?)\[\/audio\]/ism", 'tryoembed', $Text);
+	} else {
+		$Text = preg_replace("/\[video\](.*?)\[\/video\]/", '$1', $Text);
+		$Text = preg_replace("/\[audio\](.*?)\[\/audio\]/", '$1', $Text);
 	}
 
 	// html5 video and audio
 
 
-	$Text = preg_replace("/\[iframe\](.*?)\[\/iframe\]/ism", '<iframe src="$1" width="425" height="350"><a href="$1">$1</a></iframe>', $Text);
-
+	if ($tryoembed)
+		$Text = preg_replace("/\[iframe\](.*?)\[\/iframe\]/ism", '<iframe src="$1" width="425" height="350"><a href="$1">$1</a></iframe>', $Text);
+	else
+		$Text = preg_replace("/\[iframe\](.*?)\[\/iframe\]/ism", '<a href="$1">$1</a>', $Text);
 
 	// Youtube extensions
 	if ($tryoembed) {
@@ -278,7 +282,10 @@ function bbcode($Text,$preserve_nl = false, $tryoembed = true) {
 	$Text = preg_replace("/\[youtube\]https?:\/\/www.youtube.com\/embed\/(.*?)\[\/youtube\]/ism",'[youtube]$1[/youtube]',$Text); 
 	$Text = preg_replace("/\[youtube\]https?:\/\/youtu.be\/(.*?)\[\/youtube\]/ism",'[youtube]$1[/youtube]',$Text);
 
-	$Text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism", '<iframe width="425" height="350" src="http://www.youtube.com/embed/$1" frameborder="0" ></iframe>', $Text);
+	if ($tryoembed)
+		$Text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism", '<iframe width="425" height="350" src="http://www.youtube.com/embed/$1" frameborder="0" ></iframe>', $Text);
+	else
+		$Text = preg_replace("/\[youtube\]([A-Za-z0-9\-_=]+)(.*?)\[\/youtube\]/ism", "http://www.youtube.com/watch?v=$1", $Text);
 
 
 	if ($tryoembed) {
@@ -287,8 +294,12 @@ function bbcode($Text,$preserve_nl = false, $tryoembed = true) {
 	}
 
 	$Text = preg_replace("/\[vimeo\]https?:\/\/player.vimeo.com\/video\/([0-9]+)(.*?)\[\/vimeo\]/ism",'[vimeo]$1[/vimeo]',$Text); 
-	$Text = preg_replace("/\[vimeo\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/vimeo\]/ism",'[vimeo]$1[/vimeo]',$Text); 
-	$Text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism", '<iframe width="425" height="350" src="http://player.vimeo.com/video/$1" frameborder="0" ></iframe>', $Text);
+	$Text = preg_replace("/\[vimeo\]https?:\/\/vimeo.com\/([0-9]+)(.*?)\[\/vimeo\]/ism",'[vimeo]$1[/vimeo]',$Text);
+
+	if ($tryoembed)
+		$Text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism", '<iframe width="425" height="350" src="http://player.vimeo.com/video/$1" frameborder="0" ></iframe>', $Text);
+	else
+		$Text = preg_replace("/\[vimeo\]([0-9]+)(.*?)\[\/vimeo\]/ism", "http://vimeo.com/$1", $Text);
 
 //	$Text = preg_replace("/\[youtube\](.*?)\[\/youtube\]/", '<object width="425" height="350" type="application/x-shockwave-flash" data="http://www.youtube.com/v/$1" ><param name="movie" value="http://www.youtube.com/v/$1"></param><!--[if IE]><embed src="http://www.youtube.com/v/$1" type="application/x-shockwave-flash" width="425" height="350" /><![endif]--></object>', $Text);
 
@@ -327,6 +338,23 @@ function bbcode($Text,$preserve_nl = false, $tryoembed = true) {
 	if(strlen($saved_image))
 		$Text = str_replace('[$#saved_image#$]','<img src="' . $saved_image .'" alt="' . t('Image/photo') . '" />',$Text);
 
+	// Clean up the HTML by loading and saving the HTML with the DOM
+	// Only do it when it has to be done - for performance reasons
+	if (!$tryoembed) {
+		$doc = new DOMDocument();
+		$doc->preserveWhiteSpace = false;
+
+		$Text = mb_convert_encoding($Text, 'HTML-ENTITIES', "UTF-8");
+
+		$doctype = '<!DOCTYPE html PUBLIC "-//W3C//DTD HTML 4.0 Transitional//EN" "http://www.w3.org/TR/REC-html40/loose.dtd">';
+		@$doc->loadHTML($doctype."<html><body>".$Text."</body></html>");
+
+		$Text = $doc->saveHTML();
+		$Text = str_replace(array("<html><body>", "</body></html>", $doctype), array("", "", ""), $Text);
+
+		$Text = str_replace('<br></li>','</li>', $Text);
+	}
+
 	call_hooks('bbcode',$Text);
 
 	return $Text;
diff --git a/include/items.php b/include/items.php
index 87d6ab4794..6ab681c875 100755
--- a/include/items.php
+++ b/include/items.php
@@ -352,7 +352,7 @@ function get_atom_elements($feed,$item) {
 					$res['author-avatar'] = unxmlify($link['attribs']['']['href']);
 			}
 		}
-	}			
+	}
 
 	$rawactor = $item->get_item_tags(NAMESPACE_ACTIVITY, 'actor');
 
@@ -384,7 +384,7 @@ function get_atom_elements($feed,$item) {
 						$res['author-avatar'] = unxmlify($link['attribs']['']['href']);
 				}
 			}
-		}			
+		}
 
 		$rawactor = $feed->get_feed_tags(NAMESPACE_ACTIVITY, 'subject');
 
@@ -409,7 +409,7 @@ function get_atom_elements($feed,$item) {
 		$res['app'] = strip_tags(unxmlify($apps[0]['attribs']['']['source']));
 		if($res['app'] === 'web')
 			$res['app'] = 'OStatus';
-	}		   
+	}
 
 	// base64 encoded json structure representing Diaspora signature
 
@@ -553,7 +553,7 @@ function get_atom_elements($feed,$item) {
 
 		foreach($base as $link) {
 			if(!x($res, 'owner-avatar') || !$res['owner-avatar']) {
-				if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar')			
+				if($link['attribs']['']['rel'] === 'photo' || $link['attribs']['']['rel'] === 'avatar')
 					$res['owner-avatar'] = unxmlify($link['attribs']['']['href']);
 			}
 		}
@@ -697,6 +697,11 @@ function get_atom_elements($feed,$item) {
 
 	call_hooks('parse_atom', $arr);
 
+	//if (($res["title"] != "") or (strpos($res["body"], "RT @") > 0)) {
+	//	$debugfile = tempnam("/home/ike/log", "item-res2-");
+	//	file_put_contents($debugfile, serialize($res));
+	//}
+
 	return $res;
 }
 
@@ -1644,7 +1649,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $pass = 0)
 
 	// Now process the feed
 
-	if($feed->get_item_quantity()) {		
+	if($feed->get_item_quantity()) {
 
 		logger('consume_feed: feed item count = ' . $feed->get_item_quantity());
 
@@ -1657,7 +1662,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $pass = 0)
 
 		foreach($items as $item) {
 
-			$is_reply = false;		
+			$is_reply = false;
 			$item_id = $item->get_id();
 			$rawthread = $item->get_item_tags( NAMESPACE_THREAD,'in-reply-to');
 			if(isset($rawthread[0]['attribs']['']['ref'])) {
@@ -1671,7 +1676,7 @@ function consume_feed($xml,$importer,&$contact, &$hub, $datedir = 0, $pass = 0)
 					continue;
 
 				// Have we seen it? If not, import it.
-	
+
 				$item_id  = $item->get_id();
 				$datarray = get_atom_elements($feed,$item);
 
-- 
2.39.5